Depends how you're acquiring them - it's the same as Durable Objects. You can send them all to a sta

K

kianOP•10/30/25, 2:02 PM

https://github.com/cloudflare/templates/blob/36f193b322af58da4366becdae40936410ebda68/containers-template/src/index.ts#L44-L68

Ttao no, container versions not incrementing😳

N

nicoritschel•10/30/25, 2:07 PM

Yeah mine too sadly

Kkian Depends how you're acquiring them - it's the same as Durable Objects. You can se...

T

tao•10/30/25, 3:19 PM

async fetch(request: Request, env: Env): Promise<Response> {
    // note: "getRandom" to be replaced with latency-aware routing in the near future
    try {
      const instanceNumber = env.WORKER_ENV === 'production' ? 2 : 1;
      const containerInstance = await getRandom(
        env.API_CONTAINER,
        instanceNumber,
      );
      const rs = await containerInstance.fetch(request);
      return rs;
    } catch (error) {
      // Surface failures cleanly in Cloudflare's logs before bubbling up
      console.error('[CloudflareContainer] fetch failed', {
        url: request.url,
        method: request.method,
        message: error instanceof Error ? error.message : String(error),
        stack: error instanceof Error ? error.stack : undefined,
      });
      throw error instanceof Error ? error : new Error(String(error));
    }
  },

async fetch(request: Request, env: Env): Promise<Response> {
    // note: "getRandom" to be replaced with latency-aware routing in the near future
    try {
      const instanceNumber = env.WORKER_ENV === 'production' ? 2 : 1;
      const containerInstance = await getRandom(
        env.API_CONTAINER,
        instanceNumber,
      );
      const rs = await containerInstance.fetch(request);
      return rs;
    } catch (error) {
      // Surface failures cleanly in Cloudflare's logs before bubbling up
      console.error('[CloudflareContainer] fetch failed', {
        url: request.url,
        method: request.method,
        message: error instanceof Error ? error.message : String(error),
        stack: error instanceof Error ? error.stack : undefined,
      });
      throw error instanceof Error ? error : new Error(String(error));
    }
  },

async fetch(request: Request, env: Env): Promise<Response> {
    // note: "getRandom" to be replaced with latency-aware routing in the near future
    try {
      const instanceNumber = env.WORKER_ENV === 'production' ? 2 : 1;
      const containerInstance = await getRandom(
        env.API_CONTAINER,
        instanceNumber,
      );
      const rs = await containerInstance.fetch(request);
      return rs;
    } catch (error) {
      // Surface failures cleanly in Cloudflare's logs before bubbling up
      console.error('[CloudflareContainer] fetch failed', {
        url: request.url,
        method: request.method,
        message: error instanceof Error ? error.message : String(error),
        stack: error instanceof Error ? error.stack : undefined,
      });
      throw error instanceof Error ? error : new Error(String(error));
    }
  },

async fetch(request: Request, env: Env): Promise<Response> {
    // note: "getRandom" to be replaced with latency-aware routing in the near future
    try {
      const instanceNumber = env.WORKER_ENV === 'production' ? 2 : 1;
      const containerInstance = await getRandom(
        env.API_CONTAINER,
        instanceNumber,
      );
      const rs = await containerInstance.fetch(request);
      return rs;
    } catch (error) {
      // Surface failures cleanly in Cloudflare's logs before bubbling up
      console.error('[CloudflareContainer] fetch failed', {
        url: request.url,
        method: request.method,
        message: error instanceof Error ? error.message : String(error),
        stack: error instanceof Error ? error.stack : undefined,
      });
      throw error instanceof Error ? error : new Error(String(error));
    }
  },

This is my code — I am not using a static ID. If I switch to using a static ID, does that mean my container rollout time won’t be as long?

K

kianOP•10/30/25, 3:40 PM

Rollouts depend on how quickly your old Containers are shutting down (they should handle SIGTERM signals) & how quickly the new ones can start

<

<@235148962103951360>•10/30/25, 4:01 PM

Is there any way to set the location a container starts in?

T

tao•10/30/25, 8:07 PM

If the old containers stay active and never shut down, does that mean I won’t be able to deploy new containers?

T

tao•10/30/25, 8:26 PM

5h for updating....that's strange

Kkian Rollouts depend on how quickly your old Containers are shutting down (they shoul...

T

tao•10/30/25, 8:35 PM

If you were the user, would you accept waiting 5 hours for a new version to be redeployed? What if there’s a hot fix that needs to go live immediately? I’d like to know what we can do to reduce rollout time?

K

kianOP•10/30/25, 8:35 PM

Give us an application ID to investigate? It’s not normal behaviour but there’s no IDs for us to look at so far.

Kkian Give us an application ID to investigate? It’s not normal behaviour but there’s ...

T

tao•10/30/25, 8:37 PM

yes ofc. but how to view the appliaction ID

Kkian Give us an application ID to investigate? It’s not normal behaviour but there’s ...

T

tao•10/30/25, 8:39 PM

Application ID: a0351ab5-04ef-428b-9e03-0f45fc78fd25

K

kianOP•10/30/25, 8:45 PM

We’ll take a look - there’s a known bug that a new instance we’re trying to bring up can get stuck & we’re not correctly replacing it

T

tao•10/30/25, 8:53 PM

Thank you very much — I really appreciate you looking into this.
Hoping the fix can be rolled out as soon as possible.

K

kianOP•10/30/25, 8:54 PM

Generally we expect rollouts to take as long as old instances shutting down and starting new ones - new ones can sometimes get stuck and we need to fix some things to make sure the stuck ones are retried

K

kianOP•10/30/25, 8:54 PM

The rollout strategy options right now are only “proceed to the next step once X% of instances are on the new version” but we’ll bring more options over time

Kkian We’ll take a look - there’s a known bug that a new instance we’re trying to brin...

T

tao•10/31/25, 2:42 AM

could you please let us know when the fix is expected to be completed?
Currently each deployment takes a long time and it’s causing impact.

T

tao•10/31/25, 3:43 AM

We’re currently unable to deploy a new version because it’s stuck in the updating state

<

<@235148962103951360>•10/31/25, 3:43 AM

Can the location of a container be set manually?

J

Jadugar_Jaggu•10/31/25, 7:39 AM

Cloudflare should remove egress to its own products like kv and r2 from a cloudflare container.

Its preventing me to use it for a real world purpose.

Good for toying.

JJadugar_Jaggu Cloudflare should remove egress to its own products like kv and r2 from a cloudf...

1

1984 Ford Laser•10/31/25, 7:44 AM

They've said that they plan on doing this in the future. I imagine that will happen when they have bindings within the container

<<@235148962103951360>Can the location of a container be set manually?

1

1984 Ford Laser•10/31/25, 7:48 AM

Nah

<

<@235148962103951360>•10/31/25, 10:19 AM

Well that sucks. It would be great to be able to choose the location.

T

tao•10/31/25, 10:26 AM

And I also noticed another issue — even when the status is still updating and the container version hasn’t increased, new containers still start and run.
This is quite confusing, because it’s hard to tell whether my new version has actually been deployed successfully or not.

<<@235148962103951360>Can the location of a container be set manually?

J

John Spurlock•10/31/25, 4:21 PM

yes, you can constrain a container to run in a given city (EWR etc) or region (ENAM etc)

JJohn Spurlock yes, you can constrain a container to run in a given city (EWR etc) or region (E...

<

<@235148962103951360>•10/31/25, 5:05 PM

How would I set that? I tried looking in the docs and couldnt find anything about it other than it saying the containers automatically choose a location.

<<@235148962103951360>How would I set that? I tried looking in the docs and couldnt find anything abou...

J

John Spurlock•10/31/25, 5:07 PM

it's there in the API. I'm not sure how to do it in wrangler, maybe look here for clues? https://github.com/cloudflare/workers-sdk/blob/9c5601a7f5dc11869bf29fe5b4cb5c4558a75951/packages/workers-utils/src/config/environment.ts#L192

GitHub

workers-sdk/packages/workers-utils/src/config/environment.ts at 9c5...

Home to Wrangler, the CLI for Cloudflare Workers® - cloudflare/workers-sdk

J

John Spurlock•10/31/25, 5:08 PM

it definately does work though, I just reconfirmed both region and city constraints this morning on new container apps

JJohn Spurlock it's there in the API. I'm not sure how to do it in wrangler, maybe look here fo...

1

1984 Ford Laser•10/31/25, 7:57 PM

This definitely needs to be documented somewhere

JJohn Spurlock it definately does work though, I just reconfirmed both region and city constrai...

<

<@235148962103951360>•11/1/25, 4:20 AM

Thanks ill try that.

11984 Ford Laser This definitely needs to be documented somewhere

G

gruntlord6•11/1/25, 5:12 AM

agreed, I get its a beta but this kind of stuff needs to be documented well

JJohn Spurlock it's there in the API. I'm not sure how to do it in wrangler, maybe look here fo...

<

<@235148962103951360>•11/1/25, 7:59 AM

  "containers": [
                {
                    "class_name": "TestContainer",
                    "image": "./src/durableObjects/containers/TestContainer/Dockerfile",
                    "max_instances": 10,
                    "instance_type": "standard-1",
                    "rollout_active_grace_period": 0,
                    "rollout_step_percentage": [
                        100
                    ],
                    "constraints": {
                        "regions": ["OC"]
                    }
                }
            ],

  "containers": [
                {
                    "class_name": "TestContainer",
                    "image": "./src/durableObjects/containers/TestContainer/Dockerfile",
                    "max_instances": 10,
                    "instance_type": "standard-1",
                    "rollout_active_grace_period": 0,
                    "rollout_step_percentage": [
                        100
                    ],
                    "constraints": {
                        "regions": ["OC"]
                    }
                }
            ],

  "containers": [
                {
                    "class_name": "TestContainer",
                    "image": "./src/durableObjects/containers/TestContainer/Dockerfile",
                    "max_instances": 10,
                    "instance_type": "standard-1",
                    "rollout_active_grace_period": 0,
                    "rollout_step_percentage": [
                        100
                    ],
                    "constraints": {
                        "regions": ["OC"]
                    }
                }
            ],

  "containers": [
                {
                    "class_name": "TestContainer",
                    "image": "./src/durableObjects/containers/TestContainer/Dockerfile",
                    "max_instances": 10,
                    "instance_type": "standard-1",
                    "rollout_active_grace_period": 0,
                    "rollout_step_percentage": [
                        100
                    ],
                    "constraints": {
                        "regions": ["OC"]
                    }
                }
            ],

this is my config. i ran a deploy but the container still says the last location it was in, which isn't in the region i specified. I also get a warning in the wrangler config saying: Property constraints is not allowed

Wrangler also doesnt show the edit in the list of applied changes it shows when deploying

edit: seems i had to delete the container completely to get it to start in the right location

Eeve0415 How long shoud I wait?

<

<@235148962103951360>•11/1/25, 8:04 AM

mine was done after i went to bed and woke up

<

<@235148962103951360>•11/1/25, 8:04 AM

many hours

M

Minori•11/1/25, 9:43 AM

so I was thinking of being able to use Containers as a backend for our game servers, so let's say our matchmaking code that runs on workers or DO, we wanna spawn a gameserver that remains as long as the session is active or there's players in it, is that possible?

MMinori so I was thinking of being able to use Containers as a backend for our game serv...

<

<@235148962103951360>•11/1/25, 10:04 AM

it is possible to an extent

<

<@235148962103951360>•11/1/25, 10:04 AM

as far as i know, the containers dont expose ports and arent really accessible publicly

<

<@235148962103951360>•11/1/25, 10:04 AM

however, there are ways around it such as using playit.gg or an ssh tunnel to a machine/instance that is able to expose the ports etc

<

<@235148962103951360>•11/1/25, 10:05 AM

i have run a minecraft server with containers and it works

<

<@235148962103951360>•11/1/25, 10:05 AM

disk is ephemeral, so youll have to backup/restore game server data/files

<

<@235148962103951360>•11/1/25, 10:06 AM

https://github.com/eastlondoner/mineflare

GitHub

GitHub - eastlondoner/mineflare

Contribute to eastlondoner/mineflare development by creating an account on GitHub.

M

Minori•11/1/25, 10:06 AM

We really don't mind the ephemeral nature of Containers as these instances are meant to elastically spin up and down so their player progress is stored elsewhere

<

<@235148962103951360>•11/1/25, 10:07 AM

the only thing is just exposing your game server

<

<@235148962103951360>•11/1/25, 10:07 AM

playit.gg or an ssh tunnel are 2 of the options

M

Minori•11/1/25, 10:07 AM

Gotcha thanks

<<@235148962103951360>https://github.com/eastlondoner/mineflare

<

<@235148962103951360>•11/1/25, 10:08 AM

have a look at this it should give u an idea of how you could do it

<

<@235148962103951360>•11/1/25, 10:08 AM

the only issue is the instance specs

<

<@235148962103951360>•11/1/25, 10:09 AM

you need an enterprise plan to customise them i believe, or ur stuck with the default presets which arent exactly ideal for game servers e.g. 1/2 vcpu and 4gb ram for minecraft for example isnt great. the ram is ok for a small survival sever but 1/2 cpus is too weak

<

<@235148962103951360>•11/1/25, 10:09 AM

i tried with 1/2 cpu/4gb and loading chunks took a while

Depends how you're acquiring them - it's the same as Durable Objects. You can send them all to a sta

Similar Threads

Similar Threads

Similar Threads